Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
rat Risk of HIV Transmission from Blood Transfusion The risk (R) of receiving a blood transfusion containing HIV depends on a number of factors, including the prevalence (P) of HIV viremia in the donor population, the likelihood (L) of donation during the preantibody phase of viremia, the sensitivity (S) of the ELISA screening test to detect HIV antibody when antibody is in fact present, and the number (n) of units of blood (or blood products) received. The risk (D) that a donated unit of blood contains HIV can be expressed as: D = PL + P(1-LO-S). (1) The first part of the expression to the right of the equals sign, PL, represents the portion of risk due to blood donation during the preantibody phase of infection. The remainder of the equation, P(1 - L) (1 - S), represents the portion of risk due to donors with antibody whose blood escapes detection by the screening test. The pertinent concern from the vantage point of a blood recipient is the following: Given the number of units transfused, what is the risk (R) that one or more units contain HIV? This is approached by first asking the computationally simpler, related question: What is the likelihood that none of the transfused units contains HIV? If n units are transfused, the likelihood (C) that none contains virus can be expressed as: C = `1 -D)n 309 . (2)
310 APPENDIX C TABLE C-1 Definitions of Variables Entering Estimates of Risk of HIV Transmission from Blood Transfusion Symbol Definition tl Probability of HIV viremia in a donor Mean time between development of HIV viremia and antibody formation Mean time between development of HIV viremia and development of clinical indications of disease (i.e., duration of possible unknowing . . transmission) £ Relative likelihood of blood donation soon after HIV exposure compared with later donation L D C n R Likelihood of donation during the preantibody phase of viremia Risk that a donated unit of blood contains HIV Likelihood that no unit of blood contains HIV Number of units of blood transfused into a blood recipient Risk of exposure to HIV from n units of transfused blood Then the risk, R. of exposure to HIV in an individual receiving transfused blood is: R= 1-C. (3) Each of the elements entering Equation 1 is uncertain, though some data are available to provide a basis for estimates. Tables C-1 and C-2 summarize the definitions, baseline assumptions, and optimistic and pessimistic ranges in assumptions for each variable. Following are comments on the variables entering the equations: P: The nationwide experience with blood donor screening suggests that the frequency of HIV in the United States donor population may be between 1 and 10 per 10,000 (Schorr et al., 1985~. Lower figures may TABLE C-2 Assumptions in the Estimation of Risk of HIV Transmission from Blood Transfusion Assumptions and Calculations Very Very SymbolBaseline Optimistic Optimistic Pessimistic Pessimistic P0.0004 0.0002 0.0001 0.001 0.001 to (weeks)4 4 4 4 4 to (weeks)260 410 520 260 130 £ 1.0 0.75 0.50 1.0 1.0 S 0.99 0.995 0.995 0.985 0.98
APPENDIX C 3 ~ ~ prevail in low-prevalence areas and also may be realized through in- creased efforts to discourage donors who have engaged in high-risk behaviors and by eliminating those who previously tested positive for HIV antibodies. In areas with a higher prevalence of lIIV infection, the frequency of HIV among donors can be expected to be higher. Also, as more heterosexuals become infected (e.g., the female partners of men who are bisexual or intravenous ding users), the frequency of HIV infection may increase among donors who have no known history of risk behaviors. L: The likelihood of donation by an infected person during the preantibody phase depends on the average duration of time stir in which the patient may be viremic prior to formation of detectable antibody, the average length of time (t2) during which a blood donor may be unknow- in~lv infectious (i.e.. the interval between viremia and development of O , .. . ~ . ~ . ~ ~ ~ 1 ~1 1_~- _ 1-1 _1-~_~ rams _~ 1~_~:~ clinical indications of disease), and the relative ~e~noou tt~ or aonaung blood soon after exposure. Specifically: L = £t~lt2. (4) Evidence from a small number of patients suggests that the mean interval between exposure and antibody formation is approximately eight weeks (Lancet, 1984; Tucker et al., 19851. Though hard data are lacking, it is reasonable to assume that viremia is extremely unlikely during the first third of this interval, moderately likely during the second third, and almost surely present during the final third. The net result would be an effective average period of viremia prior to antibody formation of approx- imately four weeks stir. The duration of time that a donor may unknow- ingly be an HIV carrier is speculative. The baseline estimate for this mean interval time (t2) between development of viremia and development of .. . . .. . clinical disease Is 5 years (Jaffe et al., 1985), with a range from 2.5 to 10 years. The likelihood of donating blood soon after exposure may be lower than the likelihood of donating later because, for example, an isolated homosexual encounter may be forgotten or discounted in memory over time. A countervailing tendency may be for some to donate soon after an exposure to reassure themselves about their freedom from disease. The baseline relative likelihood (£) of early versus later donation is 1.0 (equal likelihood). The optimistic range extends down to 0.5; the calculations based on pessimistic assumptions retain the baseline value of 1.0 for the relative likelihood (hi. S.: Practically no diagnostic test is perfect. The sensitivity of the ELISA screening test refers to its ability to detect antibody when antibody is present. (The test's specificity, or its ability to exclude antibody when none is present, is another measure of performance, though one no relevant to this analysis.) Measurements of test sensitivity require an
312 APPENDIX C TABLE C-3 Risk of Exposure to HIV from Blood Transfusion (expressed as 1 per nearest thousandth) Number of Units Very Very Transfused Baseline Optimistic Optimistic Pessimistic Pessimistic 1 1/99,000 1/407,000 1/1,133,000 1/33,000 1/20,000 2 1/50,000 1/204,000 1/566,000 1/17,000 1/10,000 2.9a 1/34,000 1/140,000 1/391,000 1/11,000 1/7,000 3 1/33,0P0 1/136,000 1/378,000 1/11,000 1/7,000 4 1/25,000 1/102,000 1/283,000 1/8,000 1/5,000 5 1/20,000 1/81,000 1/227,000 1/7,000 1/4,000 6 1/17,000 1/68,000 1/189,000 1/6,000 1/3,000 7 1/14,000 1/58,000 1/162,000 1/5,000 1/3,000 8 1/12,000 1/51,000 1/142,000 1/4,000 1/2,000 9 1/11,000 1/45,000 1/126,000 1/4,000 1/2,000 10 1/10,000 1/41,000 1/113,000 1/3,000 1/2,000 15 1/7,000 1/27,000 1/76,000 1/2,000 < 1/1,000 20 1/5,000 1/20,000 1/57,000 1/2,000 < 1/1,000 30 1/3,000 1/14,000 1/38,000 < 1/1,000 < 1/1,000 aThe approximate average number of units transfused, based on the total number of units of blood used annually in the United States (10 million) divided by the average number of patients receiving blood in a year (3.5 million). independent truth standard that could demonstrate unequivocally the presence of antibody. No such absolute standard exists; hence, estimates of sensitivity cannot be regarded as firmly established. Various manufac- turers have provided estimates of ELISA test sensitivities ranging from 0.95 to 0.996 (Reesink et al., 1986; Weiss et al., 19861. The baseline sensitivity for this analysis is 0.99, with a range from 0.98 to 0.995. n: The number of units of blood received in transfusion depends, of course, on individual clinical circumstances. The average number of units received by a patient may be approximated by dividing the total number of units of blood used annually in the United States (10 million) by the number of patients receiving blood during a year (3.5 million). This figure of approximately 2.9 units per patient is probably a slight underestimation because a single patient receiving blood on separate hospital admissions would be counted as two separate patients in the year's tabulations. Table C-3 summarizes the results of the risk of exposure to HIV for recipients of varying numbers of units of transfused blood. The results are shown using baseline assumptions and using the range of optimistic and pessimistic assumptions taken from Table C-2. The risk to a hypothetical blood recipient from an average number of units transfused is, under baseline assumptions, approximately 1 in 34,000.
APPENDIX C 313 Progressively more optimistic assumptions yield estimates of approxi- mately l in 140,000 and 1 in 391,000. Progressively more pessimistic assumptions produce estimates of approximately 1 in 11,000 and 1 in 7,000. The risk rises as the number of units transfused increases. For example, under baseline assumptions the risk from a single unit of blood is approximately 1 in 99,000; from 5 units the risk is 1 in 20,000; and from 10 units the risk is 1 in 10,000. The recipient of 4 units under optimistic assumptions has about the same risk (1 in 102,000) as does the recipient of 1 unit under baseline assumptions (1 in 99,0001. The recipient of 1 unit under pessimistic assumptions has about the same risk (1 in 33,000) as does the recipient of 3 units under baseline assumptions (1 in 33,0001. The introduction of antibody testing has reduced the risk of HIV transmission to blood recipients by a substantial margin. In other words, instead of nearly 4,000 recipients of viremic units, the baseline calcula- tions project the transfusion of only about 100 (3,500,000/34,000) HIV- infected units each year. Most of the current risk from HIV in blood transfusion relates to the possibility of blood donation during the preantibody phase of HIV infection. This emphasizes the importance of self-selection by potential donors to eliminate those who have engaged in high-risk behaviors. The wide range in the estimated risk under different sets of assumptions highlights the uncertainty in these projections. As experience and addi- tional data become available, more accurate estimates of risk may be possible. REFERENCES Jaffe, H. W., W. W. Darrow, D. F. Echenberg, P. M. O'Malley, J. P. Getchell, V. S. Kalyanaraman, R. H. Byers, D. P. Drennan, E. H. Braff, J. W. Curran, and D. P. Francis. 1985. The acquired immunodeficiency syndrome in a cohort of homosexual men. A six-year follow-up study. Ann. Intern. Med. 103:210-214. Lancer. 1984. Needlestick transmission of HTLV-III from a patient infected in Africa. Lancet II:1376-1377. Reesink, H. W., J. G. Huisman, M. Gonsalves, et al. 1986. Evaluation of six enzyme immunoassays for antibody against human immunodeficiency virus. Lancet II:483-486. Schorr, J. B., A. Berkowitz, P. D. Cumming, A. Katz, and S. G. Sandier. 1985. Prevalence of HTLV-III antibody in American blood donors. N. Engl. J. Med. 313:384-385. Tucker, J., C. A. Ludham, A. Craig, et al. 1985. HTLV-III infection associated with glandular fever like illness in a hemophiliac. Lancet II:585. Weiss, S. H., J. J. Goedert, M. G. Sarngadharan, et al. 1985. Screening test for HTLV-III (AIDS-agent) antibodies: Specificity, sensitivity, and applications. JAMA 253:221-225.